CT Scan Image Adjudication System for Threat Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In electromagnetic scanning systems, operators face challenges in efficiently identifying and classifying threat items within cluttered CT scan images of baggage items, leading to time-consuming and labor-intensive processes with a high risk of errors due to visually obstructive distributions and unclassifiable object images.
Innovation Solution
A computer-based system for adjudicating object images in CT scan data, which digitally unpacks images, classifies them into threat, benign, and unknown categories, and prompts operators for input to resolve unknowns, using multiple classifiers and iterative updates to refine classifications and reduce operator workload.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If operators manually inspect and classify all objects in CT scan images, then threat identification accuracy can be maintained, but operator time and labor increase significantly
Solution Approach 1:
The system segments the inspection task by dividing objects into different classification categories (known clear, known alarm, known clearable, known suspect, unclassifiable) and processes them differently. Automated classification handles routine objects while operators focus only on ambiguous cases, reducing overall inspection time while maintaining accuracy.
Solution Approach 2:
The patent introduces an automated classification system as an intermediary between the raw CT scan images and operator review. This intermediary pre-processes images, assigns preliminary classifications, and filters out obvious cases, allowing operators to concentrate on uncertain classifications and improving efficiency without sacrificing accuracy.
2Reliability
If operators review every object image in cluttered baggage, then all threats can be detected, but operator workload and error risk increase
Solution Approach 1:
The system extracts and removes obviously benign objects (known clear) and clearly threatening objects (known alarm) from the operator's review queue. By taking out these definitive cases, the system reduces operator workload and focuses attention on the ambiguous cases that require human judgment, thereby maintaining detection completeness while easing operational burden.
Solution Approach 2:
The system implements feedback loops where operator classifications of unclassifiable objects are used to update and improve the automated classification algorithms. This continuous learning process enhances the system's ability to automatically handle previously ambiguous cases, gradually reducing operator workload while maintaining or improving detection completeness.
3Productivity
If the system classifies all objects automatically, then operator time is reduced, but classification accuracy decreases for complex cases
Solution Approach 1:
The classification system is dynamic rather than static. It adjusts its behavior based on object characteristics, assigning different confidence levels and requiring different levels of human review. Objects with high confidence automated classifications are processed quickly, while low-confidence cases are escalated to operators, optimizing both speed and accuracy adaptively.
Solution Approach 2:
The automated classification system serves as an intermediary that handles high-volume, straightforward cases rapidly, while serving as a filter that presents only uncertain cases to human operators. This intermediary role allows the system to maximize automated processing speed for suitable cases while preserving human accuracy judgment for complex cases.
4Loss of information
If the system displays all object images to operators, then complete information is provided, but image clutter reduces identification efficiency
Solution Approach 1:
The system extracts and removes definitively classified objects (known clear and known alarm) from the operator display. By taking out these unnecessary images, the system reduces visual clutter and allows operators to focus on unclassifiable objects that require their expertise, thereby improving identification efficiency without losing critical information.
Solution Approach 2:
The display system applies local quality differentiation by presenting different information densities to different operators or for different object types. Operators reviewing unclassifiable cases receive detailed, focused information, while the overall system maintains complete information availability in the background for reference, optimizing the local display quality for each operational context.
Data Source
AI summary
An improvement to automatic classifying of threat level of objects in CT scan images of container content, methods include automatic identification of non-classifiable threat level object images, and displaying on a display of an operator a de-cluttered image, to improve operator efficiency. The decluttered image includes, as subject images, the non-classifiable threat level object images. Improvement to resolution of non-classifiable threat objects includes computer-directed prompts for the operator to enter information regarding the subject image and, based on same, identifying the object type. Improvement to automatic classifying of threat levels includes incremental updating the classifying, using the determined object type and the threat level of the object type.